This document provides guidance for AI coding agents (Claude Code, Cursor, Copilot, etc.) working with this repository.
A collection of skills for AI coding agents. Skills are packaged instructions and scripts that extend agent capabilities.
skills/{skill-name}/
├── SKILL.md # Required - Agent instructions
├── reference/ # Optional - Supporting documentation
│ └── *.md
└── scripts/ # Optional - Helper automation
└── *.sh
- Directories: kebab-case (e.g.,
payload,react-best-practices) - Main file:
SKILL.md(uppercase, exact filename) - Scripts: kebab-case with
.shextension - Reference docs: UPPERCASE with
.mdextension
Each skill file must include:
- Frontmatter with name and description
- Overview section explaining purpose
- Quick Reference for common tasks (optional)
- Essential Patterns with code examples
- Reference Documentation links (if applicable)
Example structure:
---
name: skill-name
description: When to use this skill. Trigger phrases and use cases.
---
# Skill Title
Brief description of what this skill covers.
## Quick Reference
| Task | Solution | Details |
| ---- | ---------- | ------------------------------------------------------ |
| Do X | `method()` | [REFERENCE.md#section](reference/REFERENCE.md#section) |
## Essential Patterns
### Pattern Name
\`\`\`ts
// Code example
\`\`\`
## Reference Documentation
- **[REFERENCE.md](reference/REFERENCE.md)** - Description- Keep SKILL.md under 500 lines
- Write specific, trigger-phrase-inclusive descriptions
- Use progressive disclosure via reference files
- Prefer scripts over inline code for complex operations
- Link to supporting files one level deep
The description frontmatter field is critical for skill discovery. Include:
- Primary use cases
- Trigger phrases users might say
- Technology/framework names
- Problem types the skill addresses
Example:
description: Use when working with Payload projects (payload.config.ts, collections, fields, hooks, access control). Use when debugging validation errors, security issues, relationship queries, transactions, or hook behavior.Reference files should:
- Focus on one topic each
- Include code examples with comments
- Use consistent heading structure
- Link back to related sections
If including scripts:
#!/bin/bash
set -e # Fail fast
# Status messages to stderr
echo "Processing..." >&2
# Machine-readable output to stdout
echo '{"status": "success"}'Copy skill to ~/.claude/skills/:
npx add-skill payloadcms/skillsAdd SKILL.md content to project knowledge or paste directly.
Payload development guidelines covering:
- Collections, fields, hooks, access control
- TypeScript patterns and type safety
- Security pitfalls and best practices
- Query operators (Local/REST/GraphQL)
- Plugin development
Location: skills/payload/